Hastening the Revolution
By Bob Iannucci, Ph.D., Distinguished Service Professor, ECE and Director, CyLab Mobility Research Center
at Carnegie Mellon University
The CyLab Mobility Research Center at Carnegie Mellon University was initially created to explore the transformative impact of wireless broadband communication with smartphones. Today, the Mobility Research Center’s current focus is on connected embedded systems, including applications such as smart cities and smart environments.
We actively combine graduate education, including sponsored real-world class projects, with research focused on four key problems related to connected embedded systems, like:
- Computing at ultra-low power;
- Providing time synchronization across large sensor systems;
- Mechanisms for federating separately-built large-scale systems,
- and making programming of million-node systems as easy as writing a smartphone app.
In the course of that work, we have identified low-power, wide-area networking (LPWAN) as a potentially disruptive technology. Traditional approaches based on cellular connectivity enjoy pervasive coverage and efficient channel utilization, but come at a high price. For such large-scale systems, the cost of installing and maintaining devices (sensors) could easily make such systems infeasible.
For example, we believe in the “OHIO” principle –Only Handle It Once. That is to say, we can’t afford to go around replacing millions of batteries and re-flashing firmware. Devices in such systems have to be self-powered over lifetimes on the order of 5-10 years and have to be remotely maintainable over-the-air. Cellular networks impose a fairly high energy tax on such devices. But LPWANs potentially provide the means to avoid this energy tax, doubling (or more) lifetimes based on the energy cost of communication.
Disruptive technologies are often seen as a poor substitute for an incumbent technology that, over time, eat away more and more of the applications of the incumbent. We are looking at LPWANs as disruptive, and our research aims at systematically eliminating what are seen as limitations relative to cellular in order to radically widen system applicability.
We started working with MultiTech back in 2015. With a couple of mDots and a gateway, we deployed a simple LoRaWAN network in the City of Palo Alto and sought to understand the RF propagation characteristics of LoRa technology. Our research group developed a LoRaWAN mapping application based on an mDot connected to an Android smartphone running our software. We collected received signal strength and signal-to-noise measurements, logged them in a database, and plotted signal strength maps. We also began teaching LPWAN concepts and ran student lab exercises that involved them doing drive- and walk-testing using our Android app. This served the purpose of developing their understanding of RF propagation and, importantly, gave us some ground-truth data about how well LoRa performed.
Since then, we've further developed this capability. We have found that being able to model and predict RF performance for real-world LoRaWAN networks is a significant enabler for new applications. Most would-be users of LoRa technology lack the necessary expertise and are sometimes disappointed with the performance of networks they have deployed. We’ve been very successful in using our tools to achieve excellent, robust performance of LoRa networks through a combination of mapping (using our tools) and analytical methods. By cross-connecting mathematical prediction with selective measurements, we can make informed decisions about antenna types, station placement, and gateway placement.
We partnered with the US Geological Survey (USGS) and proposed to them what we call the cyber-geophysical system: a total redesign of the USGS sensor networks to be based on LoRa technology, using some of the technologies we've developed along with solid, commercial equipment. Under their sponsorship, we have developed a line of LoRa technology-based devices called EnviSense. EnviSense, and our new follow-on project called CommonSense, provide a highly programmable (including over-the-air) framework for flexible environmental sensing.
We’ve already deployed a limited number of devices at Beale Air Force Base and at the Pepperwood Preserve – both aimed at monitoring bodies of water. Water monitoring in California and elsewhere is a vitally important application domain that is well-suited to LoRaWAN. One gateway can cover a sizable area, and long-lifetime non-contact and immersion-probe-based sensors can deliver real-time measurements. Importantly, we have developed tools for remote programming that can be used by domain experts like hydrologists to perform intelligent, distributed, adaptive sensing.
Looking ahead, many high-impact, disruptive technologies begin by solving one specific problem really well (such as environmental monitoring of water resources) and, with success, they evolve to cover an expanding base of related applications. We are very optimistic about LoRa technolgoy as a potential disruptor, and the application domains we are engaged in may well serve to be exemplars that can then be broadened and generalized.
History illustrates that in the early stages of technology development, it is essential to deliver whole-system, so-called “vertical” solutions. Our approach follows this line of thinking in that we are solving problems across the system (device design, firmware design, software design, network design, visualization design). Our application partner, the USGS, appreciates this whole-system approach.
At CMU, we have launched our so-called Urban Networking Initiative (UNI) as a way to bring together researchers, companies and organizations that share this whole-system vision and seek to work collaboratively to accelerate the transition of LoRa (and related) technologies from vertical niches to industry-changing, flexible commercial offerings.
We invite anyone interested to reach out to us and to consider joining UNI to hasten this revolution.